As enterprises embrace a more virtualised infrastructure and moving to the cloud, they have to plan carefully to avoid technical consequences. In this tip, technology expert Hamish Macarthur highlights the issues that must be addressed when growing your virtual infrastructure.
Virtualisation is a technological innovation that continues to bring real benefits to businesses. It gives users more productive usage from their system resources, and that feeds directly into the key management goals of containing or reducing business and system costs in demanding times.
Virtualisation works across all elements of the system infrastructure. Here are some examples:
- Server virtualisation decouples the operating system and application from the hardware platform. Each application is encapsulated with its operating system. This enables applications to run on almost any server platform.
- Multiple applications can run on a single physical server as if each had control of the hardware resources. This consolidation of resources is seeing server utilisation increase from levels of less than 10% to 50% or more.
- Storage virtualisation enables the pooling of storage resources. When this is allied with thin provisioning and data deduplication, capital expenditures are reduced significantly. Storage utilisation increases from 25% to 70%, and with the new generation of disk arrays, companies are finding it possible to get closer than ever before to 100% utilisation of storage.
- It makes managing storage resources and securing data simpler and delivers significant savings in operational costs.
- Desktop virtualisation enables the standardisation of client services, which brings savings in management costs associated with maintaining software updates and supporting a large number of users. The number of software licences can be reduced, contributing to significant financial savings. With standardisation, applications are available to those who need them and not distributed to everyone.
- An additional option with desktop virtualisation is to use lower-cost thin clients such as Wyse thin and zero clients which are less expensive than traditional PCs. They require less energy to operate and, because they have a longer life, do not have to be replaced as frequently.
In all these examples, the physical estate of equipment can be consolidated and reduced resulting in lower premises costs and lower energy costs to run the IT infrastructure. Virtualisation also provides the foundation for cloud services and enables the flexibility to run applications on different physical devices in different locations.
Planning your growing virtual infrastructure
To ensure there is a predictable transition to a more virtualised infrastructure, some planning is prudent. Without appropriate considerations in place, unexpected consequences may occur that will impact system performance and service levels.
Here are some issues that must be addressed as virtualisation is adopted:
- Service levels cannot be compromised, and running too many applications on a single physical server may exhaust the physical server’s computing resources. While some “oversubscription” is typical, it can compromise response times, access to storage devices and so on.
- Connectivity of servers and storage resources needs to be carefully considered. Server and network performance has increased consistently over time, while storage arrays are limited by disk latency. Applications that have a high demand on disk accesses have the data spread across disks to reduce the latency to the application. The penalty is that disk utilisation quickly recedes to levels between 10% and 30%. Storage architectures such as iSCSI, FCoE or poor provisioning of network virtualisation can exacerbate storage issues.
- Implementing desktop virtualisation can lead to stresses on the system architecture with many virtual desktop instances taxing the servers, network connectivity and storage access.
- With cloud-based services, the network connectivity must be trusted and assured, which often requires some form of redundant connectivity to the cloud provider. The choice of cloud provider must also take into account recovery services and the time frame to recover systems.
As for security, no longer is it appropriate to run virus checks and firewalls as if each virtual machine (VM) was an encapsulated physical server. Security needs to work across the VMs so that there is a holistic approach to running within a trusted and secure environment.
The issue is not that we all have to go back to a green-field site, re-architect all our systems and try and justify a total new level of expenditure. The issue is to understand from the outset that implementing virtualisation is a journey. Benefits will be realised along every step of the journey. But virtualisation administrators must recognise the knock-on implications of each step of their virtual infrastructure plan for the move to the cloud to go smoothly.
Here are some actions that professionals can take when managing the migration to and implementation of virtualised solutions, including the move to the cloud:
- Have a clear view of your system architecture, how it operates, where applications are deployed and how they are upgraded. Clear documentation and experience pay real dividends here, so senior IT administrators should lead the move to the cloud and virtualisation.
- Carefully consider how service levels will be delivered after identifying the elements of the system architecture that will be upgraded and the applications that will be supported. Look at system components in new ways -- for example buying storage on the basis of IOPs per $/€/£. To meet the response times and agreed upon service levels, understand how the applications create server loads and how this translates to demands on network connections and the output of storage devices.
- Take a staged approach. Many organisations have moved through the test phase of implementing virtualisation and are deploying it for less critical processes before embarking on business critical applications. This phased approach reduces risk and ensures that all system elements are orchestrated prior to wider deployment.
- Security, data protection and recovery processes must be consistent with the service levels committed to the business. Ensure that these critical disciplines can be applied consistently across the system infrastructure.
- Use system management tools to discover the scale of your system architecture, the servers, storage arrays, network connections and applications. Also look for tools that will assist you in monitoring activities across the architecture, and identify where challenges exist and how best to address them. Remember: Tools must be virtualisation-aware in order to provide accurate and reliable data; otherwise those tools must be upgraded.
- Vendors make great claims, but ask them to demonstrate how the claims will benefit your specific organisation. Get the vendors to commit to reducing IT expenditure whilst maintaining services. Hands-on testing and proof-of-principle projects prior to deployment will refine skill sets and help identify problem areas. This will put professionals’ strategic choices to the test going forward and paint a much clearer view of what the next steps will be.
Short-term perspectives and “we will get by somehow” approaches have failed in the past, and that will not change.
Virtualisation is a journey that will evolve to give you improved outcomes for the business. Be aware of the implications of upgrading your virtual infrastructure. Set achievable goals and a strategic direction for moving to the cloud so IT can continue to deliver for the business.
Hamish Macarthur is the founder of Macarthur Stroud International, a research and consulting organisation specialising in the technology markets. Email: firstname.lastname@example.org.